文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Object Recognition for Video Retrieval
侵权投诉

Object Recognition for Video Retrieval

Abstract. Recognition of objects in video can offer significant benefits to video retrieval including automatic annotation and content based queries based on the object characteristics. This paper describes our preliminary work toward recognizing objects i

Object Recognition for Video Retrieval

Rene Visser, Nicu Sebe, and Erwin Bakker

LIACS Media Lab, Leiden University

Niels Bohrweg 1, 2300 CA Leiden, The Netherlands

Abstract. Recognition of objects in video can offer significant benefits

to video retrieval including automatic annotation and content based

queries based on the object characteristics. This paper describes our

preliminary work toward recognizing objects in video sequences and

gives a brief survey of the relevant research in the literature. We use the

Kalman filter to obtain segmented blobs from the video, classify the

blobs using the probability ratio test, and apply several different

temporal filtering methods, which results in sequential classification

methods over the video sequence containing the blob. Results from real

video sequences are shown.

1 Introduction

Detecting particular objects in video is an important stop toward semanticunderstanding of visual imagery. For example, in content based retrieval, the ability todetect people and automobiles gives the option of advanced queries such as "Find avideo clip which contains a crowded area or a fast moving car."

In this paper, we give a summary of the current work in video segmentation formoving objects followed by the implemented tracking and classification algorithm.The novel aspect of our work is described in Sections 4 and 5 where we use temporalfiltering for classifying the moving objects and show some early results. Conclusionsare given Section 6.

2 Background

The segmentation of moving objects is an important problem in image sequenceanalysis and in the problem of video retrieval (i.e. [2, 5, 13]). There is significantrelated research [1-18]. Some approaches toward visual matching apply stochasticparsing [14]; integrate learning [1, 5, 8]; and adaptive background mixtures [7].Areal-time system for tracking people, called Pfinder (“Person finder”), is proposed byWren, et al. [16]. First, a model of the scene is built by observing the scene when noperson is present. For each pixel, the mean color value and the covariance of theassociated distribution is determined. Then when a person enters the scene, the systemM. S. Lew, N. Sebe, and J. P. Eakins (Eds.): CIVR 2002, LNCS 2383, pp. 262-270, 2002.

Springer-Verlag Berlin Heidelberg 2002

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email